Macromolecules 1994,27, 5428-5439
5428
A Synchrotron SAXS Study of Structure Development Kinetics during the Reactive Processing of Flexible Polyurethane Foam Michael J. Elwell, Stephen Mortimer, a n d Anthony J. Ryan' Manchester Materials Science Centre, UMIST, Grosvenor Street, Manchester MI 7HS, United Kingdom Received March 1, 1994" ABSTRACT: The kinetics of microphase separation during the processing of flexible polyurethane foam have been investigated. Forced-adiabatic,time-resolved synchrotron SAXS experiments were employed to probe the evolution of structure. Microphase separation was observed to occur at a critical conversion of isocyanate functional groups and shown to follow the kinetics associated with spinodal decomposition. The isocyanate conversion at the microphase separationtransition (MST)was in good agreement with our previously reported FT-IRresults. From the scattering data, R(q), the amplification rate of the composition fluctuations was determined. The data have been analyzed in terms of a time-dependentGinzburg-landau model (TDGL). Plots of R(q)/q2versus q2 exhibited a maximum at a finite value of scattering vector (4). These observations were in qualitative agreement with the theoretical predictions of the TDGL theory. Introduction The reactive processing of water-blown flexible polyurethane foam from liquid monomers and oligomers involves a complex combination of both chemical and physical events. In less than 5 min, a liquid mixture of relatively low molecular weight components is transformed into the supramolecular architecture of solid foam. Information regarding both the reaction kinetics and development of morphology during processing is essential, such that an objective description of the events taking place and, ultimately, selective control of the process can be achieved. Flexible polyurethane foam is formed by the simultaneous reaction of a diisocyanate with a polyether poly01 and water. Combination of these two exothermic reactions leads to the formation of a segmented block copoly(urethane-urea), of the -(HmS),,- type. This is blown into a foam by the cogeneration of carbon dioxide gas evolved from the water-isocyanate reaction. As the polymerization proceeds, the core of the rising foam bun becomes self-insulated by the surrounding polymer and this has the effect of bringing the process to occur under quasi-adiabatic conditions. Reaction kinetic studies during foam formation with both toluene diisocyanate (TDI) and methylene diphenyl diisocyanate (MDI) have been conducted previously and the results are documented in the literat~re.'-~ Analyses- of the final morphology present in flexible polyurethane foams employing small angle X-ray scattering (SAXS), dynamic mechanical spectroscopy (DMS), and differential scanning calorimetry (DSC) have shown them to exhibit a microphase-separated morphology similar to segmented urethane elastomers. The development of morphology during foaming is ~ o m p l e x . ~As, ~the chemical reactions proceed, the chain lengths ( N , degree of polymerization) of all the products increase and the interaction parameters (x)can also change. Such changes can give rise to the system crossing thermodynamic boundaries which results in a transition from an initial homogeneous (disordered) state into a microphaseseparated (ordered) ~tate.~JO The resultant morphology is determined by the kinetic competition between polymerization and microphase s e p a r a t i ~ n . ' ~ ~ J ~
* To whom all correspondence should be addressed.
* Abstract published in Advance ACS Abstracts, August 15,1994. 0024-929719412227-5428$04.50/0
Scattering studies of the phase separation kinetics in multiblock copolymers are not common. To our knowledge there have been only five detailed reports p~blished.1l-l~ The first two pertain to styrene-butadiene-styrene triblock" and styrene-isoprene diblock systems,12respectively. The third by Ryan and co-workers'3 involved the study of the kinetics of phase separation by SAXS during the bulk polymerization of a copolyurethane. The hard segment was composed of MDI and butane-l,4-diol, and the soft segment was a poly(ethy1ene oxide), tipped poly(propylene oxide) diol of molar mass 2000 g/mol. The two remaining reports by Chu and co-workerd4J5involved the study of the phase separation kinetics of copolyurethane samples which were quenched from the homogeneous melt phase to different annealing temperatures. The hard segments were composed of MDI and butane-lP-diol. The molecular connectivity in block copolymersrestricts the spatial extent of the concentration fluctuations to dimensions of approximately twice the radius of gyration (R,) of the entire block chain (=200A). As a consequence, probing the phase separation kinetics calls for SAXS or SANS (small angle neutron scattering). In this paper we present results that have been obtained by employing the synchrotron SAXStechnique to investigate the microphase separation behavior of water-blown flexible polyurethane foam, based on MDI and a polyether polyol, monitored under forced-adiabatic conditions. A comparison of the results obtained is made with our earlier results from forced-adiabatic, time-resolved FT-IR spectroscopy.4 The scattering data are analyzed by a generalized timedependent Ginzburg-Landau model of microphase separation kinetics proposed by Hashimoto16 and previously employed by Connell and co-workers.lZ Theory Hashimoto16 has proposed that the time variation of the order parameter, the difference between the average concentration in the homogeneous phase and the local concentration of the same component, in a block copolymer system which has undergone a quench to a lower temperature (or from a quench provided by an increase in xN)from an initially homogeneous state is given by a time-dependent Ginzburg-Landau theory. It has been shown16that the general form of the variation in scattered intensity I ( q , t ) ,with time at fixed q following a quench is given by the following equation: 0 1994 American Chemical Society
SAXS Study of Flexible PU Foam 6429
Macromolecules, Vol. 27, No. 19, 1994
The variation in I(q) at a given time interval is determined by the scattering law, P(q), in the homogeneous state. R(q) is termed the growth rate constant and is given by eq 2, where M is the mobility term, G is the Gibbs free energy, R(q) = -Mq2[
5+
2kq2]
and lz is a gradient free-energy term. For the experiments that will be discussed here, the original derivation of Leiblerl’ is most appropriate, i.e. the weak segregation limit. For an AB diblock copolymer with the average composition of the A block being ~ J A the , final expression for P(q) is given by
(3) where N is the degree of polymerization, x is the FloryHuggins interaction parameter between the component blocks, and F ( x ) is an expansion in terms of the Debye function for the scattering from a single block copolymer molecule, and composition (in terms of volume fraction), where x = q2R2 and R, is the radius of gyration of the whole block copolymer molecule. At the spinodal point, eq 3 diverges (Le. -P(q)-l = 0) and is no longer a true description of the scattering a t x L xs (2’ ITs), where T, is the spinodal temperature. The divergence and its variation with composition will define the phase boundary for the block copolymer. Modifications to eq 1 have been discussed previously by CookI8 which take into account random thermal fluctuations (inclusion of a Brownian motion term). In employing eq 1to analyze the data, the extremums are not strictly true. If R(q) is positive, then at t = m, I(q,m) is infinite; conversely, if R(q) were negative, then the scattered intensity, I(q,m), would be zero a t infinite time following a quench. Furthermore, the q dependence of the Onsager coefficient, Lo, relating the diffusive flux of copolymer molecules to the local chemical potential has been neglected. This may be valid for the early stages of phase separation and a shallow quench. However, the latter point is not valid for the systems discussed here where there is a large thermodynamic quench. It should be noted the LO generally does have a q dependence. This dependence has been calculated by Pincuslg for a polymer blend (LO 0: q-2) but not for a block copolymer. The important parameter in determining the position of the reaction mixture with respect to the phase boundary is the product xN. If the product xN < (xN),, then the value of R(q) is negative. Neglecting the Onsager coefficient, LO,R(q)/q2 can be taken as a measure of the thermodynamic driving force for the growth of the concentration fluctuation with wave vector q / 2 ~ . A negative value indicates that such a concentration fluctuation will not grow but decay away. Thus, the system remains stable to concentration fluctuations of this particular wave vector. However, a different situation prevails if (xN)> (xN)~.There is a region of q in which R(q), and thus R(q)/q2are positive and the concentration fluctuations do not decay but grow and give rise to microphase separation. These growing concentration fluctuations have upper and lower critical boundaries to their wavenumbers. Outside these limits, the concentration fluctuations decay and do not contribute to the phase separation dynamics.12 As originally published,20 the Cahn-Hilliard theory of spinodal decomposition is a
macroscopic description and has no direct relation to events at the molecular level. Extensionsto the theory have been made by de Gennes,21 Pincus,19 and Binder for polymer blends.z2 The thermodynamic driving force for the growth of the concentration fluctuation with wave vector q / 2 ~ R(q)lq2, , becomes a maximum at q = qm=. Thus, the wavelength, q / 2 ~of, the dominant Fourier component of the growing fluctuations in the early stages of phase separation is determined by the Fourier component that exhibits the maximum thermodynamic driving force. qmaxis time independent in the early stages of phase separation and is controlled by thermodynamics. R(qm,) is further controlled by the transport properties. (4)
The effective diffusion coefficient,Deff,can be determined from the value of q where maximum scattering intensity occurs, qm=,during phase separation using eq 4. Binderz2 discusses equations that are identical in form to the CahnHilliard equations, but they are couched in terms of the diffusion components of the homopolymer blend.
Experimental Section To study the polymerizationof flexible polyurethanefoam via SAXS is difficult. The material undergoes an exotherm of 75150 “C,the viscosity of the reaction medium increases from = l O to lo4Pa s, and the density decreases from lo00 kg m3 to 30 kg m3 in under 4 min. Translated into volume, this means that the final foam volume is some 33 times greater than the initial volume. Figure 1 shows typical experimental data for a flexible foam formulation prepared with 4.19 g of water/100.0g of polyol. The reaction is highly exothermic, and the amount of material required to achieve self-insulation cannot be contained within a cell of the type previously employed by Ryan and co-workers13in an optical bench assembly on a synchrotron beamline. Ideally, the temperature of the SAXS sampling cell should be identical to that of the bulk foam throughout the reaction exotherm. If not, and the cell temperature is lower than that of the foam material in contact with the cell wall, the cell will effectively act as a heat sink and, as a consequence, will decrease the foam reaction exotherm,delay the reaction chemistry,and disrupt the resulting development of morphology. Thus, rapid heating of the cell is necessary in order to be able to replicate the reaction exotherm of the foam. Also, due to the rapid rate of the foaming reaction, it is necessary to employ reaction injection moulding (RIM) to meter, mix, and inject the reaction mixture into the cell. To circumvent the heat-loss problems, a new cell has been designed which can be positioned in the optical bench assembly of a beamline and fed with a reaction mixture from a micro-RIM ma~hine.~31231~ SAXS Cell. The temperature-controlledcell was constructed from aluminum and has an internal volume of =6.6 cm3. It comprised two outer plates fitted with thermocouplesand having countersunk holes of 4 mm diameter covered with polyimide (Kapton) windows of nominal thickness 10 gm. An aluminum insert of thickness4.2 mm formed the “mold”. The cross-sectional area of the mold is some 18 times that of the runner and gate assembly. The relatively high viscosity and the low flow rate provide for laminar flow of the material in the region of the windows. A top unit, which provided extra clamping capability to the two outer sections and the mold, was fitted with sockets such that the cylindrical heating elementa could be fitted firmly in place. A Swagelock pipe was used so that the reaction mixture could be fed from the RIM machine nozzle via copper tubing (internal diameter of 1.68 mm and a typical length of 70 cm) to the SAXS cell mold. The cell itself was mounted on an X-Y translator for alignmentin the X-raybeam. The cell was equipped with four, 250 W high-densitycartridge heaters (GodfreyThermal Limited,U.K.), one positioned at each comer of the cell. A small “chimney“of aluminum foil was fitted to the top piece of the cell
Macromolecules, Vol. 27, No. 19, 1994
5430 Elwell et al.
t
c
20
-1-J 0
t
,
,
,
*
50
100 time i s
"
-+
I
I50
200
250
made from 15 wm mica; the exit window of the vacuum chamber is made from 10pm Kapton. A beamstop is mounted just before the exit window to prevent the direct beam from hitting the detector. The camera is equipped with a multiwire quadrant detector. The quadrant detector has an opening angle of 70' and an active length of 0.2 m. This detector measures intensity in the radial direction. It has an advantage over single wire detectors in that the active area increases radially. The detector can handle count rates up to approximately 250 000 counts s-l. For the calibration of the sample to detector distance, the scattering pattern from an oriented specimen of wet collagen (rat-tail tendon) was used. Parallel plate ionization counters were positioned before and after the SAXS cell and recorded the incident and transmitted intensities. Thus, changes in the attenuation factor of the specimen (transmission and thickness) resulting from the density changes during the foaming process could be monitored continuously. The experimental data obtained were corrected for background scattering (subtraction of the scattering from the empty cell and camera), sample thickness and transmission, and the positional alinearity of the detector. A schematic diagram that shows the complete experimental arrangement that was employed is illustrated in Figure 3. The SAXSpatterns were recordedevery2s. The temperature of both the cell and the reactingfoam were recorded at a frequency of 1Hz over a period of 512 s. For each run, the shot volume was typically 12.5 cm3 and the shot time was 0.42 s. The material at the window was approximately 0.05 s old. Finally, a 5 mm cube section of foam was removed from the chimney section after 30 min and a static SAXS pattern of the "final morphology" was obtained.
Results -. 0
---r-
0
__ 40
-
'
,
I
---I
"
80
120
160
200
m e is
Figure 1. (a) Adiabatic temperature rise profile and (b) the rapid change in density and elastic shear modulus during the foaming reaction for the foaming system RIMSAXS-419. to contain the excess foam and prevent the foaming mixture flowing over the top of the cell and making contact with the heating elements. A schematic diagram of the complete cell, equipped with foil chimney, cartridge heaters and type J thermocouples is given in Figure 2. Temperature Control of the Cell. The temperature of the small angle X-ray scattering cell was controlled by a Macintosh I1 microcomputer equipped with Strawberry Tree Incorporated Workbench V3.1 system software and ACM2-12-8 (T51) data acquisitionA-Dhardware. The adiabatic temperaturerise profile was prerecorded and played back as the set-point on a PID controller. Further details are documented elsewhere.s Materials. Model flexible foam formulations with water concentrations of 2.10 and 4.19 g/lOO.O g of poly01were employed throughout the work. Details of the formulations are provided in Table 1. In all the discussions that follow, the formulation containing 2.10 g of water/100.0 g of poly01will be referred to as RIMSAXS-210 and that containing 4.19 g of water, RIMSAXS419. Micro-RIM Machine. In order to be able to meter, mix, and inject a reactant stream into the cell such that accurate control of stoichiometry was maintained, a micro-RIM machine was employed. The machine has been designed to be portable to interface with analytical in~truments*3.~ and is of such a size that it could be brought into the experimental hutch without difficulty. Synchrotron SAXS. SAXS measurements were made on beamline 8.2 at the Synchrotron Radiation Source (SRS)at the SERC Daresbury laboratory, Warrington, U.K. With the SRS operating at 2 GeV and 200 mA, a flux of 4 X 1010 photons s-1 is generated at the sample position. A vacuum chamber is positioned between the sample and detector in order to reduce air scattering and absorption. Both the exit window of the beamline and the entrance window of the vacuum chamber are
Temperature Control of the Cell. A comparison of the temperature profiles for the reference temperature, the thermal response of the cell, and the foam within the cell are illustrated in Figure 4 for both RIMSAXS-210 and RIMSAXS-419 foaming systems. For RIMSAXS210 there is an initial short delay in the response of the cell of approximately 5-10 s. This results from the finite time required for relaying the signal around the circuit and the initial response of the heating elements. In the case of RIMSAXS-419, this initial delay is approximately 5 s longer. RIMSAXS-419 has a greater thermal lag than RIMSAXS-210 because its increased water concentration results in a more rapid rate of temperature rise. This demands a much faster response from the heating elements within the cell. What is important is that the cell does not act as a heat sink for the foam within the cell. The difference in temperature between the reference and the cell will clearly have an effect on the value of the isocyanate conversion determined. The error in the isocyanate conversion resulting from the difference in temperature between the reference and the cell is shown in the inset caption of Figure 4.5 The upper curve represents RIMSAXS-419 and the lower curve, RIMSAXS-210. Tref is the reference temperature at time t and Tceu is the temperature of the cell at any time t. The error in the isocyanate conversion at the onset of microphase separation ( t = 120 f 2 s) and the physical gel point (t = 180 s) is within 1% for RIMSAXS-210. In the case of RIMSAXS419, the maximum error in PNCO is approximately 5% at the onset of microphase separation (t = 60 f 2 s) and the error reduces to approximately 1% at physical gelation ( t = 80 f 2 s). Correlation between the Foam Density Calculated from Volume Rise Profiles and That Calculated from the Intensity of the Rear IonizationCounter. Parallel plate ionization counters were positioned before and after the SAXS cell, and these recorded the incident and transmitted intensities. This allowed changes in the attenuation factor of the specimen (transmission and
SAXS Study of Flexible PU Foam 5431
Macromolecules, Vol. 27, No. 19, 1994
10 deleclor
-
To RIM Machine xhead
Figure 2. 3-D schematic diagram of the SAXS sampling cell. Table 1. Details of the Foam Formulation Components formulation component polyether polyol isocyanate distilled water tertiary amine catalyst silicone surfactant
foaming system R1MSAXS-210 R1MSAXS-419 mass/g mass/g 100.00 100.00 37.10 67.66 4.19 2.10 0.70 1.80 4.00 4.00
thickness) resulting from the change in sample density during the foaming process to be monitored continuously. Figure 5 is a plot of the intensity from the rear ionization counter response during the reaction for a typical data set obtained from RIMSAXS-210. The chanee in foam density was calculated by normalization of t i e intensity change exponentially (since intensity is proportional to e-p). This was then scaled with the initial density of the reaction mixture. Reasonable agreement between the densitycalculatedas afunctionoftime fromthe absorption of the X-rays and that from foam volume expansion profiles measured by ultrasonic methods'J was observed. In addition, the final foam density observed in the SAXS experiments was in close agreement with that obtained from laboratory bench-scale foaming. Figure 6 compares the density change observed from the volume rise profiles with that determined from the absorption of the X-rays. Synchrotron Small X-ray Scattering. Representative time-resolved SAXS data that have been collected during foam formation are presented in Figures 7 and 8 for RIMSAXS-210 and RIMSAXS-419, respectively. They are three-dimensional plots of intensityZ(q,t), versus scattering vector, q. versus time, t. The first frame (an
empty cell) has been subtracted from the subsequent patterns t o remove the background (cell and camera) scattering. The data have also been corrected for changes in transmission (due to the density change) and the positional alinearity of the detector. It will be observed that for both systems, there is an initial upturn in 1(q)at low q. It is proposed that this is due t o the filling of the cell with a liquid that contains air bubbles which are growing in size. The air bubbles are of approximately 1-10 pm in diameter at the start of the reaction. It is thermodynamically more favorable for the carbon dioxide which is evolved from the water-isocyanate reaction to undergodiffusionintotheexistingbubblesthantoundergo self-nucleation to evolveanew bubble under theconditions of foam formati0n.~5 It should be emuhasized that the number of bubblesremains approxima&yconstant.26The air bubbles (which are introduced by mechanical agitation prior to loading of the polyol blend into the tank of the RIM machine) act as the nucleation sites and grow due to the diffusion of carbon dioxide from the water-isocyanate reaction. At the endofthe reaction, themean cell diameter is on the order of 20(t600 pm.
Thepatternsforbotbsystemsillustratethatintheearly stage of the reaction, there is a homogeneous liquid present. In the case of RIMSAXS-210, after 120 2 s (see Figure 12a), there is the first appearance of a scattering maxima and this occurs a t q a 0.06 k'. The intensity at this scattering maxima continuously increases until approximately 200 8, after which, the growth in the peak intensity slows down. Beyond 220 8, the peak intensity becomes approximately constant. This is after theexpanding foam has reached the Berghmans pointz6(onset of vitrification, p ~ = 0.71 ~ o 0.02). In the case of RIMSAXS-419, after
*
*
Macromolecules, Vol. 27,No. 19, I994
5432 Elwell et al.
Reactant
X-ray source h=1.54A
for feeding the heated SAXS cell
-
Heated aluminium SAXS Cell equipped
each corner
comprising the hending mirrors and Ge crystal monochromator onnected via a relay circuit to Maclntosh I1 Personal Computer
central data acquisition system Figure 3. Schematic diagram illustrating the complete experimental arrangement within the X-ray hutch.
*
60 2 8 (refer to Figure 12h), there is the first appearance of a scattering maxima and this occurs at q = 0.05 A-1. The intensity a t this scattering maxima continuously increases until approximately 80 2 s, after which the growth of the peak intensity slows down. Beyond 96 2 s, the peak intensity becomes approximately constant. As with RIMSAXS-210, this is after the vitrification @NCO = 0.71 0.02) of the expanding foam. The longer induction time prior to the appearance of the scattering maxima in RIMSAXS-210 results from the lower rate of reaction of this system compared with that of the higher water concentrationsystem,RIMSAXS-419. Suchohservations were reproducible on a run to run basis (X5 runs) for the two systems investigated. Interdomain Spacing. The maximum inl(q) suggests the presence of structure with periodic electron density within the sample. The most common practice for determiningthe periodicity is to report the Bragg spacing, d, as given by eq 5. Figure 9 is a plot of I ( q ) versus q a t
*
*
*
d = 2?r/q,,
ideal electron density variance, ( q 2 d (electrons2 em4), of a perfectly separated two-phase system with sharp phase boundaries is given by eq 6. Where q is the average electron
(5)
selected times, for arepresentativedataset for the foaming system RIMSAXS-419. The maximum in Z(q) occurs at q = 0.05 A-', yielding an interdomain spacing of 126 A. This was reproducible to within *2 A on a run to run basis. For the foaming system, RIMSAXS-210, the interdomain spacing was calculated t o he 105A. This was reproducible on a run to run basis to within *3 A. The d spacing did not change during the structuring process for either system. Degree of Microphase Separation. The growth in scattered intensity during the polymerization can he related to the degree of microphase separation through the square of the electron density variance, (&). The
density of the material, and @ and go are the volume fraction and electron density of the two pure phases, respectively. A real system will contain thermal density fluctuations and diffuse boundaries between the phases. In this case, the actual electron densities are closer to the average, leading to a decrease in the electron density variance of the real system which is given by eq 7. The (72)
= $1(11- r1'Y
+ M a - %)2
(7)
valueof ( q 2 )will thusvary between zerofor a homogeneous mixture and ( q 2 0 ) for a fully microphase-separated block copolymer. Experimentally, the electron density variance may he calculated from eq 8. Where i, is the Thompson
scattering factor and the quantity Q is known as the invariant. It is known as the invariant because it is independentofthe sizeor spatialarrangementof structural heterogeneities. The invariant is a linear function of the electron density variance, ( q 2 ) , and a quadratic function The experimental invariant of the volume fraction, can he employed to characterize the structural development as well as the degree of microphase separation. Determination of the electron density variance, ( q 2 ) , requires absolute invariant data which in turn require absolute intensity measurements. The absolute intensity
+.
SAXS Study of Flexible PU Foam 5433
Macromolecules, Vol. 27,No.19, 1994 I , , , , I , , , , I , , , , I , , , , I , , , , I , , . , I
L
825
80
30
130
180
230:
time Is
550
t
27s
